The Google Cloud PPA: six levers that decide what the aggregate discount is worth
A Private Pricing Agreement is not a discount you accept. It is an aggregate spend commitment, a flat percentage layered over your CUDs, and a set of clauses that decide whether the commitment protects your budget or quietly traps it. The headline percentage is the smallest part of the deal.
Prepared by Redress Compliance · June 2026 · Representative Google Cloud estate scenario (benchmark scenario, not a quote)
Executive summary
A Google Cloud Private Pricing Agreement, the PPA, is a negotiated commercial overlay on the standard Google Cloud customer agreement. You commit to an aggregate spend floor over a one, three, or five year term.
In exchange you receive a flat percentage off list across covered services. That percentage layers on top of any Committed Use Discount and sustained use discount you already hold.
The flat aggregate discount is modest. On an annual commit of $5M to $15M, the PPA overlay typically lands at 12 to 18 percent above standard rates. The deep discounting still comes from the CUDs underneath, which reach up to 55 percent on steady compute. The PPA stacks on top of that, it does not replace it.
The number Google will not volunteer is the cost of the commitment itself. The aggregate floor is a take or pay number. Under the consumption model effective July 15, 2025, an unmet commitment is billed and gone, so an inflated aggregate is a direct loss, not a recoverable credit.
On a representative $18.0M annual estate, a baseline anchored program built from CUDs, a right sized PPA overlay, Marketplace pull through, and egress credits cuts run rate to about $11.07M, a saving of roughly $6.9M or about 38 percent. The leverage is in the six levers and five clauses below, not in chasing the largest aggregate commit.
What is a Google Cloud PPA and how is it priced in 2026?
A PPA is a custom pricing addendum to the standard Google Cloud customer agreement. You promise an aggregate spend over a term, and Google grants a flat discount across covered services plus, in larger deals, a credit pool, migration funding, and service specific deal pricing.
Two things sit in one document, and buyers conflate them at their cost. The first is the aggregate spend commitment, a take or pay floor. The second is the flat discount you get for signing it. The discount is the reward; the commitment is the risk.
The aggregate discount scales with the size of the annual commit. The public bands seen across 2026 deals run as below, with strategic accounts negotiating above the top tier.
Chart A. Indicative flat PPA overlay by annual commitment tier, applied above standard and CUD rates. Strategic accounts negotiate above the top band.
Lever one: how big should the aggregate commitment be?
The aggregate floor is the whole game. A PPA sized to a Google forecast is sized to Google's interest. A PPA sized to your verified consumption floor is sized to yours.
Build the floor from twelve months of billing export, mapped to services and regions. Commit the steady base, never the peak, and never the growth case the account team draws on the whiteboard.
How to build a baseline that survives scrutiny
Google will challenge a low baseline with its own usage telemetry. Your defense is the billing export, not an estimate. Three rules from the engagement file hold up under that pressure.
- Anchor to verified billing: use the detailed billing export, not a forecast deck, and commit between the p25 and p50 of steady monthly spend.
- Strip one off spikes: remove migration bursts, load tests, and seasonal peaks from the committable floor.
- Stage the term: put the stable base on the longer term and keep the variable layer on a shorter commit or on demand.
Lever two: how does the Vertex AI overlay change the PPA?
Negotiating BigQuery and Vertex AI spend into the PPA, rather than paying on demand, produces savings in the 20 to 40 percent band on eligible usage. AI spend is the fastest growing and least predictable line in most 2026 estates, which makes it the line to handle with care.
Vertex AI list pricing is per token and public. Gemini 2.5 Flash Lite lists at about $0.10 per million input tokens and $0.40 per million output, while Gemini 2.5 Pro starts near $1.25 input and $10.00 output per million.
As of Google Cloud Next 2026, Vertex AI is consolidating into the Gemini Enterprise Agent Platform. Pin your terms to the spend, not the product name.
Commit the floor, not the forecast
Treat the AI commit as a hedge on the inference floor, the part that runs in production every day. Keep training and experimentation on demand. A spend based commit discounts the dollar, not the token rate, so a mid term model price cut flows to you only if your clause pins the discount to net spend at then current rates.
Lever three: how does Marketplace pull through retire the commitment?
Google Cloud Marketplace third party purchases retire your aggregate commitment at up to 100 percent of the purchase value. That is the single biggest structural advantage Google holds over AWS, which caps most Marketplace spend at 25 percent of the commitment.
The mechanic is published in the Marketplace private offer documentation. Software you would buy anyway, routed through Marketplace, becomes commitment attainment at full value.
Use Marketplace as a relief valve on the aggregate floor, not as a reason to inflate it. The two are different decisions: the floor is your risk, the Marketplace pull through is your insurance against missing it.
Lever four: how do egress credits fit the PPA?
Egress is rarely discounted inside the CUD or the flat PPA percentage, so it has to be negotiated as a separate credit or commit line. It is the cost that most often surprises a finance team at renewal, and it is the cost vendors use to raise switching friction.
Two facts give the buyer leverage in 2026. Both belong in the PPA negotiation, not after it.
- Free exit transfers: Google offers free data transfer out when a customer fully exits Google Cloud, a response to regulatory pressure on lock in. Confirm the terms and the process in writing.
- EU Data Act: switching and egress provisions under the EU Data Act remove the worst of the egress penalty for in scope customers, which weakens the lock in argument the account team relies on.
Ask for an egress credit pool sized to your real cross region and outbound volume, and pin the per gigabyte rate for the term so a mid contract rate card change cannot erode it.
Lever five: which five contract clauses protect the commitment?
The discount is one line. The five clauses below are what stand between you and a year end true up. Negotiate them before you trade the discount, because Google concedes terms more easily while the headline percentage is still open.
| Clause | What it protects | Buyer side ask |
|---|---|---|
| Net spend definition | What counts toward the aggregate floor | Include CUD spend, Marketplace pull through, and credits in attainment |
| True down | Relief when the estate shrinks | Defined reduction points at each anniversary |
| Shortfall recovery | Grace if you miss the floor | Carry forward window before any true up is billed |
| Renewal rate | Protection against renewing at list | Renew at the original negotiated rate, not then current list |
| Exit and portability | Unused commit and data on exit | Credit or portability of unconsumed commit, free exit egress |
The net spend definition is the clause that pays
The most valuable clause is the dullest. If the order form defines net spend to exclude Marketplace pull through or CUD spend from the aggregate attainment, your floor is far harder to reach than it looks. Pin a broad net spend definition first, then negotiate the percentage.
The auto renewal trap
PPAs renew automatically at term end, and many default that renewal to then current list rather than your negotiated rate, with a short cancellation window. Put the renewal rate and a clear cancellation notice period in writing, or you re sign your discount away by silence.
Lever six: what buyer side moves neutralize Google's tactics?
The account team runs a familiar playbook into a PPA. Each move has a counter that costs you nothing but preparation.
| Google move | What it sounds like | Buyer counter |
|---|---|---|
| Anchor to forecast | Commit your projected growth for the deepest rate | Commit the verified p25 to p50 floor only |
| Lead with the flat percentage | The PPA gets you 18 percent off everything | Model the CUD stack first, treat the PPA as the overlay |
| Five year push | A longer term unlocks the best discount and funding | Five year base only on the stable core, shorter on the rest |
| Quarter end urgency | This rate expires at period end | Use Google's quarter end as your timing leverage |
| Bundle and obscure | Roll credits, funding, and rate into one number | Itemize every line so each is separately exitable |
How do the six levers add up on a real estate?
The framework runs on a twelve week cycle: build the baseline, model the levers and clauses, then negotiate and stage. The phases below are the cycle we run on the buyer side.
Baseline
Export twelve months of detailed billing. Map spend to services and regions. Set the committable floor at the p25 to p50 of steady usage, stripped of one off spikes. Map third party renewals that can route through Marketplace.
Model and clause
Model the CUD stack first, then the PPA overlay. Size the aggregate to the verified floor. Draft the five clauses, lead with the net spend definition, and price the Vertex AI and egress lines.
Negotiate and stage
Benchmark the opening quote, hold the discount counter until the clauses close, and sign at quarter end for timing leverage. Stage the commit between a base tier and a variable tier.
Applied to a representative estate, the program looks like the table below. Each workload is committed only to its steady share, with the variable remainder left on demand.
| Workload | On demand annual list | Instrument | Effective discount | Net annual |
|---|---|---|---|---|
| Compute Engine, GKE, Cloud Run | $9.0M | CUD 3 year plus PPA overlay | 42% | $5.22M |
| BigQuery Enterprise | $4.0M | Edition commit plus PPA | 40% | $2.40M |
| Vertex AI and Gemini | $3.0M | Spend commit plus PPA overlay | 35% | $1.95M |
| Egress and other | $2.0M | Egress credit and Marketplace | 25% | $1.50M |
| Total estate | $18.0M | blended | about 38% | $11.07M |
Benchmark scenario, not a quote. Benchmark ranges: Redress Compliance advisory engagement file, 2024 to 2025.
Chart B. Representative $18.0M estate. Committed program run rate $11.07M, a $6.9M or about 38 percent reduction, with no aggregate commit above verified steady state.
What discount should you expect across renewal and exit scenarios?
Benchmarks from the engagement file frame what good looks like across scenarios. New platform commits land deepest; over committed shortfall positions land worst. The ranges below are effective total discount off list, CUDs and PPA combined.
Chart C. Effective total discount achieved by scenario. Source: Redress Compliance advisory engagement file, 2024 to 2025.
Layered above the CUD stack, not instead of it.
On eligible AI and analytics spend at the three year step.
The worked $18M estate at $11.07M, baseline anchored.
Across 25 to 35 PPA negotiations we ran in 2024 to 2025.
Where the common advice on Google Cloud PPAs is wrong
The standard reseller and account team pitch is to commit a large multi year aggregate to capture the deepest flat percentage. We disagree. Across the PPA negotiations we benchmarked in 2024 to 2025, the inflated aggregate commit was the most reliable way to overpay.
Two reasons. First, the flat PPA overlay is small, 12 to 18 percent in the common band, against CUDs that already reach far deeper. Stretching the aggregate to win a few more flat points puts a large take or pay floor at risk for a thin reward.
Second, since the July 2025 consumption model, an unmet commitment is billed and gone, not a recoverable credit. The shortfall costs more than the extra points ever saved.
The buyer side move is to size the aggregate to the verified floor, let the CUD stack and Marketplace pull through do the heavy lifting, and treat the flat PPA percentage as the overlay it is. A smaller, safer commit beats a larger one you cannot reach.
The contrarian line: the largest aggregate commit on the wrong baseline loses to a right sized one on the verified floor. The flat percentage is the smallest lever in the room.
What is your BATNA and the side letter language?
A PPA negotiation without a credible alternative is a price taker negotiation. Build the BATNA before the first call, and make sure the account team knows it exists.
- Azure: a Microsoft Azure Consumption Commitment sized to the same baseline, with reservations and savings plans underneath.
- AWS: an EDP or Savings Plans for the portable workloads, priced for comparison.
- OCI: Oracle Cloud Universal Credits for the database heavy share, where rates often undercut.
- Stay on demand: the always available alternative that caps your downside if the terms are wrong.
Side letter language we use
Pin the protections in a side letter when the standard order form will not carry them. Three clauses do most of the work.
- Net spend: define attainment to include CUD spend, Marketplace pull through, and credits.
- Renewal: renew at the original negotiated rate with a defined cancellation notice.
- Shortfall: a carry forward window before any true up is billed.
Get these in writing or treat the discount as unprotected.
Recommendation. Size the aggregate to your verified floor, win the net spend definition, then trade the flat discount.
- Do this first: build the twelve month baseline, model the CUD stack, and size the aggregate commitment between the p25 and p50 of steady spend, never the forecast.
- Do this next: close the net spend definition, true down, shortfall recovery, renewal rate, and exit clauses before you accept the headline percentage, and map third party renewals through Marketplace for full commitment attainment.
We work on the buyer side only, and we are glad to tie a meaningful part of the fee to delivered value.
Frequently asked questions
What is a Google Cloud PPA?
A Google Cloud Private Pricing Agreement is a custom commercial addendum to the standard customer agreement. You commit to an aggregate spend over a one, three, or five year term, and Google grants a flat percentage off list across covered services, layered on top of your existing CUDs and sustained use discounts.
Does the PPA replace my committed use discounts?
No. The flat PPA percentage layers on top of the CUD stack, it does not absorb it. Model the CUDs first, where the deep discounts live, then treat the PPA as the overlay. A quote that presents the PPA percentage as your total discount is hiding the CUD layer underneath.
What happens if I miss the aggregate commitment?
You still owe the floor. The aggregate commitment is take or pay, so an unmet commit is billed at term end. Since the consumption model effective July 15, 2025, that shortfall is not a recoverable credit, which is why sizing the aggregate to the verified floor matters more than the headline rate.
Can Marketplace spend reduce my PPA commitment?
Yes. Eligible Google Cloud Marketplace third party purchases retire your aggregate commitment at up to 100 percent of value, far more generous than the 25 percent cap AWS applies. Route software you would buy anyway through Marketplace to help attain the floor and avoid a shortfall true up.
How long does a PPA negotiation take?
Plan twelve weeks. Four to build the verified baseline, four to model the levers and draft the five clauses, and four to benchmark, negotiate, and stage the commit. Signing at Google's quarter end adds timing leverage.
Renewing a Google Cloud commitment?
Talk to a buyer side advisor. Thirty minutes, your consumption baseline and the five terms worth pinning, before the account team sets the commit for you.
Buyer side intelligence, monthly
One letter a month. Negotiation moves, audit signals, and price book shifts.